[DRILL-2456] regexp_replace using hex codes fails on larger JSON data sets

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.7.0
• Fix Version/s: Future
• Component/s: Functions - Drill

Environment

Drill 0.7
MapR 4.0.1
CentOS

• Assignee: Unassigned
• Reporter: Andries Engelbrecht
• Created: 13/Mar/15 23:00
• Updated: 01/Oct/15 16:37

Description

This query works with only 1 file

select regexp_replace(`text`, '[^\x20-\xad]', '°'), count(id) from dfs.twitter.`/feed/2015/03/13/17/FlumeData.1426267859699.json` group by `text` order by count(id) desc limit 10;

This one fails with multiple files

select regexp_replace(`text`, '[^\x20-\xad]', '°'), count(id) from dfs.twitter.`/feed/2015/03/13` group by `text` order by count(id) desc limit 10;

Query failed: Query failed: Failure while trying to start remote fragment, Encountered an illegal char on line 1, column 31: '' [ 43ff1aa4-4a71-455d-b817-ec5eb8d179bb on twitternode:31010 ]

Using text in regexp_replace does work for same dataset.
This query works fine on full data set.

select regexp_replace(`text`, '[^ -~¡-ÿ]', '°'), count(id) from dfs.twitter.`/feed/2015/03/13` group by `text` order by count(id) desc limit 10;

Attached snippet drillbit.log for error